OpenAI just scooped human researchers on a million-dollar quantum math problem

OpenAI said its proprietary internal language model produced a proof of the existence portion of the Yang-Mills problem, a Clay Mathematics Institute Millennium Prize Problem concerning the mathematical foundations of quantum chromodynamics. The company said the model reached the result on September 9 and that it withheld the proof while translating it into Lean, a proof-verification programming language. OpenAI has not publicly released the proof and instead sent it to outside experts for preparation and review. Yang-Mills theory describes quantum fields used in the Standard Model of particle physics. The Millennium Prize problem requires proofs that the theory is mathematically consistent and has a nonzero mass gap; OpenAI’s announced result addresses only the existence component. Mathematicians Levent Alpöge of Anthropic and Ethan Sussman of Northwestern University had independently pursued a related proof using Anthropic’s Claude models, publishing progress in a GitHub repository beginning October 6. They updated intermediate results the night before OpenAI’s announcement and believed they had solved the three-dimensional existence problem, while the Clay problem requires four dimensions. Sussman said their work could have been weeks from completion. OpenAI’s decision to keep its proof private pending Lean verification drew criticism from Alpöge, Sussman and other mathematicians, who called for clearer scientific and publication norms for AI-generated mathematics.